Skip to main content

Full text of "HENING TATI"

See other formats


THE ANCIENT LANGUAGE OF AZEEBAIJAN 1 
By W. B. Henning 

When I proposed the title of this paper to the officers of the 
Society, I imagined, in an optimistic mood, that I could offer 
a reasonable theory of the language once spoken in Azerbaijan. 
Further study, I regret to say, has convinced me that I can- 
not : our information is defective on all sides ; and the 
information we do possess is affected by uncertainty in the 
most essential points. Although therefore it would be wise to 
keep silent, I hope I may be allowed to put the problem as an 
unsolved problem, as well as to add a little to the information 
hitherto available. 

For several centuries now a form of Turkish has been the 
common language of the north-western provinces of Persia, 
Azerbaijan and Zenjan. The late Persian historian, Sayyid 
Ahmed-i Kesravl, has traced the story of the gradual infiltra- 
tion of Turkish tribes into those territories, and the attendant 
regression and final disappearance of the older population and 
their language ; the process began in the 11th century and was 
completed by the beginning of the 16th. 2 

It is generally agreed, and indeed not subject to serious 
doubt, that before the advent of the Turks Iranian languages 
were spoken here in Azerbaijan and Zenjan, as elsewhere in 
Persia. From the distribution of the Iranian dialects one may 
infer the group of Iranian to which the lost language of 
Azerbaijan belonged. To the east of Azerbaijan, in the high 
mountains that enclose the southern edge of the Caspian Sea, 
and in the coastal plain itself, we have successively Talis!, 
Gllaki, and Mazandarani, also called Tabari ; and beyond the 
mountain range, in the neighbourhood of Semnan, several 
further dialects. To the south-east of Azerbaijan, at a great 
distance, we find the Central Group of dialects in the neigh- 
bourhood of Isfahan, with branches eastwards towards Yezd 

1 Paper read before the Philological Society on Dec. 4, 1953. 

2 A. Kesravi, ASari yii zaban-i bastan-i Aharbayagan. Tehran 1304/1926. 
With regard to the alleged survival of ASari in Tabriz down to the end of 
the 16th century, see below, p. 176, n. 5. 



158 TRANSACTIONS QF THE PHILOLOGICAL SOCIETY 1954 

and southwards towards Slvand. In the same direction, at less 
distance, a group of dialects was recently noticed in the 
neighbourhood of Vafs, half-way between Hamadan and 
Save. 1 In the south, GuranI survives in the Zagros mountains, 
which separate Persia from the plains of Iraq. And in the far 
west, beyond the limits of Persia, as far as the western border 
of classical Armenia, Zaza, called Dimli by its speakers, 
existed until quite recently and for all we know may exist even 
now. All these languages, which may be said to surround 
Azerbaijan, belong solidly to the north-western group of 
Iranian, and that was probably true also of the lost Azer- 
baijanian tongue. 

The languages and dialects named just now constitute all 
the surviving forms of North-west-Iranian speech about which 
we have information, with the sole exception of Kurdish, 
which stands apart and is outside the range of the present 
subject. Several have attained full status as literary languages : 
for example GuranI, the language of an obscure religion, the 
Ahl-i Haqq, with considerable literature ; Gurgani, 2 from the 
south-eastern corner of the Caspian Sea, now defunct, once the 
language of a Muslimic sect, the Huxufl; and above all, 
Tabari, with a literary history looking back almost as far as 
the Persian literary language. 3 Nevertheless, all of them have 
been receding before the onslaught of the official language of 
the country, Persian, which itself belongs to a different group, 
to South-western Iranian ; and now, under modern conditions, 

1 M. Muqaddam, GuyiShayi Vafs ve Asliyan ve Tafris (= Iran-Kude 
Xo. II). Tehran, 1318 Yezd./1949. This work also contains notices, 
deserving attention, of a local gypsy dialect, a Southern Kurdish language 
(Zand), and Khalaj Turkish. 

2 Our knowledge of that dialect, hitherto largely based on Huart, Texles 
persons relatifs a la secte des HouroHfis (Gibb Mem. Ser. ix), 1909, has been 
deepened by S. Kiya, Vaze-name-i Gurgani (Inlisarat-i Daniigah-i Tehran 
133), Tehran 1330/1951. An interesting survival is the optative (3rd sg. 
and pi.) byndy/bndy, which corresponds with Parthian bwyndyy. [See 
now further H. Hitter, Die Anfdnge der Hurufisekte, Oriens vii, 1954, 1-54.] 

3 The older specimens were recently collected and analysed by S. Kiya 
(Vaze-name-i Tabari [Iran-Kude No. 9] Tehran, 1316 Yezd./1947), who 
also published the text of a Tabari " Nisab ". 



W. B. HENNING — THK ANCIENT LANGUAGE OF AZERBAIJAN 159 

all the non-Persian languages and dialects are rapidly dying 
away. 

Of the dialects still spoken in villages some may well dis- 
appear before they have been recorded. True, some areas have 
been fairly fully investigated ; but others have been neglected, 
most of all the north-western provinces. Not only is practically 
no information available about the few dialects known to have 
survived in remote corners of Azerbaijan itself ; but also the 
districts between that province and the capital, a stretch of 
some 250 miles, have remained disregarded, and this in spite 
of a hundred years' study of the dialects of Persia. It is 
probably due to such gaps in our knowledge that the task of 
determining the relationship of the north-western dialects to 
each other continues to present such great difficulty. 

Thanks to a lucky chance I am able to say a little about 
a dialect spoken on the approaches to Azerbaijan, roughly 
halfway between Tehran and the present border of that 
province. I came upon it 1 three years ago at Takistan, a 
village formerly known by the name of Siyah-dehan, some 
twenty miles to the south-west of Qazvin, on the ancient 
high-road that joins Azerbaijan with the heart of Northern 
Persia. I had only two hours and a half for taking notes, and 
even this short period was not free from disturbances of various 
kinds. In the outcome, the material I collected is insufficient 
and defective ; regrettably there was no chance to check 
doubtful points and fill in the gaps that became manifest as 
soon as it was possible to read over my notes at leisure. In 
fact, one would be inclined to bury them quietly, were it not 
for the evident importance of this dialect, which may con- 
veniently be called TakistanI, for the comparative study of 
North-western Iranian. 

At the time I did not know that the dialect had in fact been 
mentioned once before, namely by Professor Minorsky in the 
Encyclopaedia of Islam, in his article " Tat ". Professor 

1 I owe its acquaintance to the kindness of Mr. Hannibal, of Tehran, 
a very learned gentleman who invited me to pay a visit to Qazvin, a town 
for which he had unbounded enthusiasm. 

PHILO. TRANS. 1954. N 



160 TRANSACTIONS ■ OF THE PHILOLOGICAL SOCIETY 1954 

Minorsky there expressed the opinion that it resembled the 
dialects of the Central Group and quoted, from his own 
observation, a small number of the words he regarded as 
characteristic. Our notes are not always in agreement. For 
example, of the five verbal forms quoted by Professor Minorsky 
(mtzanti ' I know ', mizanu ' we know ', mizaninda ' they 
know ', bislcas ' look ' [imp.], si ' you should go ') two differ 
materially from the forms I received, in which the first person 
of the singular ends in -im and the first person of the plural in 
-um. The full set of the endings of the present in Takistanl is 
as follows : Sing, -im, -i, -e ; Plur. -um, -a, -inda. 

A related dialect is known from another village in the 
neighbourhood of Takistan, Istihard ; known, however, not by 
observation but through a written source. Zukovskiy found 
a copy of a Persian dictionary, the Burhdn-i J ami', which was 
printed in Tabriz in 1844 : in the margins of that copy a 
Persian had written glosses in an otherwise unknown dialect, 
attributed to Istihard. 1 These glosses, which number about 
180, are written in Arabic script, with fairly full vowel marks ; 
they are of restricted usefulness, chiefly because many of them 
merely serve to indicate a slightly deviating pronunciation of 
Persian words. Nevertheless, there is enough to show that this 
dialect is very near to Takistanl. Two words may suffice here. 
The first, titiye or tifiye ' daughter ' (Tak. titlye), is typical of 
the dialect group, with its initial t- ; the nearest form otherwise 
is MahallatI ditiye. The second, burbunistin ' to weep ' (Tak. 
birbanastan), has four characteristic points : the use of the 
prefix hi- with an infinitive ; the ending -astan ; the metathesis 
-rb- ; and -n- in the place of -m-, the base being bram-. 

Now I shall put before you a few of the points which define 
the position of Takistanl in relation to the other North- 
western Iranian languages. The first is the survival of gram- 
matical gender. I had been told that ' bull ' is gdv and ' cow ' 
magave, which in addition to the prefixed ma- (an old adjective 
meaning ' female ') possesses a short vowel at the end, which 

1 2ukovskiy, Materiali, vol. i, p. ix. The Istihardt words are included in 
the glossary in vol. ii, part i. 



W. B. HENNING — THE ANCIENT LANGUAGE OF AZERBAIJAN 161 

could be a mark of gender. Towards the end of our session 
I was blessed with an intelligent informant and drew his 
attention to the difference, and lie told me, almost in so many 
words, that his language possessed gender distinctions, and 
quoted as example, mardak biso ' the man went ', but zeiriiye 
bisiye ' the woman went '. Here we have separate forms for 
the verb, the 3rd person singular of the intransitive preterite, 
which however is a nominal form by origin. Later, when 
I looked over my notes, I found several additional examples, 1 
one of which shows that this distinction is not confined to 
words of natural gender : vara bumia ' snow fell ' (literally 
' snow came ') ; here both the noun and the verb have 
feminine endings. 

Gender distinctions exist in comparatively few of the 
North-western languages ; of those reasonably near to 
Takistan, in Semnani and in certain Central dialects, such as 
Farizandl and Josagani. 2 In the details, Takistani agrees here 
most closely with Semnani. The Semnani equivalent of the 
sentence ' snow fell ' happens to occur in the texts collected 
by Christensen, vara . . . biamia 3 ; it almost coincides with the 
Takistani version. 

There are other points, some of them almost as weighty, in 
which Takistani agrees with Semnani. Semnan is situated 
about 200 miles to the east of Takistan ; half-way between the 
two villages sprawls the modern capital of Persia, whose 
presence has extinguished the intervening dialects. 

Yet Takistani has another face, and that is turned towards 
the North. The pronominal system shows that most clearly. 
Practically all the pronominal forms of the dialect are men- 
tioned in the following table : — 
Personal 
Enclitic 
Possessive 
[Tal. 

1 iimi numaziye bisiye ' my betrothed went ' ; harhe biparasti ' the 
chicken flew (up) '. 8 Lambton, Three Persian dialects, p. 44. 

3 Le dialecte de Samnan, p. 62, line 3. 



az 


ta 


av 


ama 


suma 


ana 


-m 


-i 


-s 


-mun 


-yun 


-sun 


cimi 


istd 




Suma 


suma 


jand 


cimi 


isti 


cai 


cama 


sima 


cavon] 



162 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 

Here the possessive pronouns are by far the most interesting. 
Possessive pronouns are something of a rarity in the North- 
western Iranian languages ; they also do not exist in Persian, 
the language of communication used for talking with the 
villagers, so they attract one's attention immediately. It is 
true, they function occasionally also as oblique cases of the 
personal pronouns, as for example in the sentence cimi iUa 
dnidiyindd ' they do not give me to you ' (cimi ' me ', istd 
' you ', d(n) verbal prefix, ni negation, diyindd ' they give'). 
Such use is compatible with their origin ; for they consist 
of an ancient preposition, Middle Iranian ac ' from ', and 
oblique cases of the personal pronouns, themselves no 
longer in use. 1 But their characteristic function is that 
of possessive pronouns ; e.g. cimi sigdr ' my cigarette ', 
)d fiyar 2 kala ' her father's head ', cumd gdlbar ' our 
gate '. 

Comparable pronouns exist only in one of the North- 
western languages hitherto known, in Talis!, the language of 
Talis, the district on the Caspian Sea which straddles the 
frontier between Persia and the Soviet Union. The Talis! 
pronouns, which there are used exclusively as possessive 
pronouns, are given above ; they are in form almost identical 
with the Takistani series. 

Another proof of the close relationship between Takistani 
and TalisI is provided by the preterite of the verb. Below a 
full set of the normal forms has been given, successively the 
preterite of the intransitive verb, the preterite of the transitive 
verb, the preterite of ' to be ' in post-sonantic position, and 
the pluperfect of a transitive verb, which involves the preterite 
of ' to be ' : — 



1 Cf. the oblique cases of the pronouns for the 3rd person in Semnani, 
Christensen, p. 43. 

- piyar/piar is oblique ease of pia ' father ', ef. piarpia ' grandfather ', 
pianmpia ' my grandfather ' ,- similarly maya ' mother ', mayum ' my 
mother ', manmaya ' (my) mother's mother ' ; zanbdra ' wife's brother ', 
bararzan ' brother's wife '. Cf. Christensen, Samnan, § 80. Curious is fir 



W. B. HENNING — THE ANCIENT LANGUAGE OF AZERBAIJAN 163 



I sat 

anistim 

anistis 

anist 

anistimun 

anisteiyun 

anistinde 



I said 

vdtim 

vati 

vdtis 

vdtimun 

vdiiyun 

vdtisun 



I was 

-veime 

-veise 

-ve 

-veimun 

-veiyun 

-vinde 



I had said 

vdta-vima 

vdta-vl 

vdta-viM 

vdta-wmun 

vdtd-vlyun 

vata-vlsun 



Here the most striking feature is the threatening confusion of 
the second singular with the third singular : anistis ' you sat ', 
but vdtis ' he said ', veise ' you were ' but vdta-visa ' he had 
said '. What preserves the difference is merely the status of 
the verbs as either transitive or intransitive, but this distinc- 
tion is in process of being reduced, as one can see from the 
1st and 2nd persons of the plural, where the intransitive has 
borrowed the finals of the transitive ; one may perhaps 
wonder how the contrast between anistis ' you sat ' and vdtis 
' he said ' is going to be resolved. 

There is no difficulty about the origin of these forms. The 
intransitive, of course, consists of the perfect participle and 
the present of 'to be ', but the transitive of the perfect 
participle and the enclitic pronouns. What interests us here is 
the agreement of the intransitive with the corresponding forms 
in Talis!. In that language the present of ' to be ', in combina- 
tion with the negation (ne), has the following forms : nim, nis, 
ni (niye) ; nimon, nion, nin. There is substantia] agreement, 
and that is most remarkable in the 2nd person of the singular, 
-is ' you are '. Such a form is exceptional in Iranian. It is 
true, one finds it here and there, for example in Eastern 
Iranian in Sogdian ; and in South-western Iranian among 
some dialects of Laristan, on the coast of the Persian Gulf 1 : 
but in North-western Iranian it was hitherto known solely 
from Talis!. 

A considerable distance separates Takistan from the 
Talisi-speaking area, and the greater part of it is occupied by 



1 Mann, Tdjik-Mundarten, 127 sqq. ; Ivanow, Gabri dialect, 77 ; Romas- 
kevio, Lar i ego dialekt (Iranskie Yaziki, i, 1945), 41 et passim. 



164 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 

the most inaccessible mountain country in the whole of Persia. 
There is every likelihood that TalisI and dialects close to it 
extended much towards the south, into the mountains, prob- 
ably as far as Khalkhal and upper Tarom, 1 and we may 
assume that dialects related to Takistani extended to the 
north-west of its present location, towards the Zenjan valley, 
so that there may have been a smooth transition from Talis! 
to Takistani, just as there probably was a smooth transition, 
through dialects now lost, from Takistani to Semnani in the 
east, and to the Central dialects, or some of them, in the south. 
Thus we may regard Takistani as the essential link, joining 
the Northern, Eastern, and Southern groups. That it is in the 
right and natural position is also indicated by certain traits it 
shares with GllakI and Tabarl, the languages in its neighbour- 
hood to the north and north-east. One could mention, e.g., 
the preference for -astan as the secondary ending of the 
infinitive 2 (an ending entirely absent from Talis!) ; or the 
almost primeval word vuye ' water ', which has disappeared 
from practically all Iranian languages, but was preserved in 
Gilan, as biya? in certain geographical names. 4 



We now leave Takistan and turn to Azerbaijan proper. As 
I mentioned at the beginning, in this province, where Turkish 

1 See below. 

2 bitasastan ' shave ', biskdsdstdn ' look ', bicasdstdn ' taste ', birbandstdn 
' weep ', dgarddstan ' turn back ', uvazdstdn ' dance '. 

3 Biya-pU and Biya-pas. It is doubtful whether any of the other dialect 
forms with initial v/w may belong here, such as Sangisarl vo (Zukovskiy), 
v°o (Christensen, ii) ; Yazdi wo/vov etc. (Hadank, Khunsdr, lxxvi n. ; 
Andreas-Christensen 102 ev6v ; Ivanow wuw) certainly represents dp-. 

4 Note also luyas ' fox ' (Tab. luwds, Gozarkhoni [Ivanow, A.O., ix, 367] 
luwos) ; pild ' big, great ', pild-mus ' rat ' (Gozarkhoni pilo bowo ' grand- 
father ' ; Gilaki pile, pille ; Zaza pil, pili Hadank 163 ; often wrongly 
confused with Pers. pir ' old ' ; Dailemi name Pilesuvdr, Minorsky, Domina- 
tion des Dailamites, p. 3). A few further' interesting Tak. words may be 
briefly mentioned here : dzira ' yesterday ' ; sard ' day after to-morrow ' ; 
cilst ' boot ' ; vayd ' wedding ' ; cdlu. ' sparrow ' ; ashe ' dog ' ; guyar ' ealf ' ; 
aslf ' apple ' ; uz ' walnut ' ; simar ' straw ' ; terizgd ' hail ' ; nimarij 
' noon ' ; zdrin ' child ' (pi. zarun ; of. Kurd.) ; geisin ' plough ' (from 



W. B. HENNING — THE ANCIENT LANGUAGE OF AZERBAIJAN 165 

has been the dominating language for several centuries, a few 
islands of Iranian speech have survived in remote corners. 
Here our information is singularly defective : instead of 
knowing any of these surviving dialects, we merely know 
rumours of their existence. Three areas have been named : — 

Firstly, the Harzan-Kuh in the north-western corner of the 
province, in the ancient borderland between Armenia and 
Persia, to the north-west of Tabriz, between the northern 
shore of the Urmia lake and the River Aras (the ancient 
Araxes). In several villages here, in Harzan, Galin-qaya, 
Babra, and others, an Iranian dialect is spoken, which may 
conveniently be called HarzanI 1 ; I shall say more about it 
presently. 

Secondly : some villages in the Qaraja-day, to the north-east 
of Tabriz. Nothing is known about their language. 2 

Thirdly : several villages in Khalkhal, at the eastern fringe 
of Azerbaijan. This is precisely the area in which we should 
expect to find dialects that provide a transition from Talisi to 
Takistani, and for this reason it is much to be regretted that 
we know nothing about them. There is at least a potential 
source of information. It is understood that the late Kesravl, 
in the second edition of his well-known book on the ancient 
language of Azerbaijan, quoted a few words in one of the 
Khalkhal dialects ; but this second edition, published in 
Tehran in 1317/1938, does not seem to have reached Western 

*gav-a$in) ; isbdrz ' spleen ' ; vale' ' kidney ' (cf. Tal. vek) ; dim ' face ' ; 
mijik ' eyelash ' ; bar ' door ' ; xur ' good ' ; gujil ' small ' ; peind ' broad ' ; 
zur ' compost ' (cf. Tal. zir/zil) ; asbune yuryuri ' it thunders ' (cf. Tal. 
gurgur) ; bikatan ' to fall ', miginim ' I fall ', az mugu biginim ' I shall fall ' ; 
bifundan ' to chew ', mujilnim ' 1 chew ' ; vinddn ' to see ' ; dast usindan 
' to touch ', unsinim ' I touch ', unsindimC!) ' I touched ' ; sozavian ' to grow ', 
sozambim, sozambt, sozambe, sozumbeyum, -beyd, -birtdd ' I grow ', sozaveime 
' I grew ' (cf. Pers. subz budan) ; adiyan ' to give ', dndim, dndey, andiye, 
andeyum, fmdeyd, dndiyindd ' i give ' ; eisayinddn ' to light a fire ', eisdma- 
sinim ' I light it '. 

1 Dr. M. Navabi recently discovered and published a Persian deed dated 
in 791/1389, in which the name now generally known as Harzand is spelt 
Harzan ; accordingly, Dr. Navabi prefers HarzanI as the name of the 
language {Nasriye-i Ddniikade-i Adabiydt-i Tabriz, V, No. ii, 29-38). 

- See the additional note below, p. 177. 



166 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 

Europe 1 ; at any rate, in spite of some search, I have not been 
able to see it. 2 

Twenty years ago a promising attempt was made to fill the 
great gap in our knowledge of these dialects, by a Swiss 
linguist, Dr. Emil Baer. In 1932/33 he visited Harzan and 
Khalkhal 3 (but not the Qaraja-day), and then went on to 
study the languages of Persian Talis, of Gilan, and Mazen- 
daran. To two successive Congresses of Orientalists, at Rome 
in 1935 * and at Brussels in 1938, s he read papers on the 
material he had collected and the method he had used. At 
the latter Congress (which I could not attend) he quoted 
examples from the languages he had investigated, but these 
examples were unfortunately omitted from the report printed 
in the Proceedings. In fact, not a single word or a single form, 
out of the clearly massive material he had brought together, 
has been made accessible to the public. It has been reported 

1 See the additional note below, p. 177. 

2 Meanwhile, thanks to the kindness of a friend in Tehran, I have secured 
the 3rd edition of Kesravi's ASari (Tehran 1325/1946). It is probably an 
unchanged reprint of the 2nd edition, regrettably printed in a most un- 
satisfactory fashion, often illegible and studded with misprints. The 
specimen of a dialect (unspecified, presumably Sahrud) of Khalkhal is on 
pp. 61-62, in unvocalized Arabic script ; its contents, a brief description of 
the linguistic situation in Khalkhal, render it almost useless for our pur- 
poses ; according to it, dialects (called Tail) are spoken in the whole of 
Sahriid and in a few villages of KayaSkunan : all of them are close to Talisi. 
As far as one can see, the latter opinion is partly borne out by the specimen, 
which shows some characteristic Tal. forms (istan ' self ' ; im ' this ' ; 
antecedent genitive, sometimes in -I ; postpos. -ku ; possessive pronoun 
le [ih] ' his ') ; but there are also considerable deviations from Talisi. 
especially in the verb ( e.g. vojin [iwjti] ' they call ', gaf-zanin ' they talk ', 
bera [brh] ' it was ', sera [sr ?] ' it went '). However, one fails to detect any 
resemblance to Takistanl, in which the corresponding verbal forms are 
mafinda, zaninda, -De, and (bi)so. 

3 He worked in due villaggi remoti, ma grandi e popolati, che han conservato 
anche essi il loro idioma iranico (Atti p. 237) . . . im Tale Shahrud, einem der 
5 Mahale des Berglands Khalkhal (Actes p. 153). 

4 II metodo della geografia linguistica applicato all' investigazione dei 
dialetti iranici. Atti del xix Congresso Internationale degli orientalisti, Rome 
1938, 233-239. 

6 Zur Dialektologie Nordirans. Actes du xx" Congres Internatioiml des 
Orientalistes, Louvain, 1940, 153-157. 



\V. B. HENNING — THE ANCIENT LANGUAGE OF AZEEBAIJAN 167 

that towards the end of the war Dr. Baer was killed in 
Germany, and it was said that his material had disappeared ; 
but recently I learned from Professor Minorsky that Dr. 
Georges Eedard had succeeded in tracing his collections. 
There is now good hope that they will be published one day. 
Pending that, we must make do with what can be obtained in 
other ways. 1 

The only one of these dialects about which I can give 
information is Harzam, or more precisely, the variety of it 
spoken in the village of Galin-qaya. It was in this village that 
European scholars first became alive to the survival of Iranian 
languages in Azerbaijan. C. F. Lehmann-Haupt, who passed 
a night at Galin-qaya in 1898, noticed its peculiar dialect, and 
drew attention to it in the report on his journey which was 
published in 1910. 2 He quoted there four or five separate 
words, only two of them significant (hard ' three ', isba ' dog '), 
and a sentence of two words, ospe bindor, which he misunder- 
stood : he thought it meant ' the horses are harnessed ', 3 
while in fact it means ' tie up the horse ', ospe being singular 
and bindor the plural of the imperative. 

This is all so far made known about the language of Galin- 
qaya ; in compensation, we have seven words from the 
neighbouring village of Harzan, which the late Mirza Muham- 
mad-i Qazvlni communicated to Professor Minorsky, who 
quoted them in the Encyclopaedia article mentioned above. 
These few words suffice to show that there are differences 
between Galin-qaya and Harzan. Two of the forms supplied 
by Muhammad-i Qazvlni, together with their Persian equiva- 
lents and the corresponding Galin-qaya forms, may be 
adduced here, berend = Pers. budand : Gal. berut ; sermdu 
= Pers. sudand : Gal. serut. 

It will be clear from the foregoing remarks that I have had 

1 However, during the recent Congress of Orientalists (Cambridge, 
Aug., 1954) Professor Redard told me that only a part of Dr. Baer's collec- 
tions had survived ; the notes on the dialects of Azerbaijan have disappeared. 

2 Armenien einst undjetzt, i, 185 sq. 

3 Die Pferde sind ' ffebunden ', angeschirrt. 



168 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 

access to a fresh source of information, and this is happily 
a copious one. 1 I owe it to a Persian scholar, Dr. M. Navabi, 
a lecturer in the recently founded University of Tabriz, the 
capital of Azerbaijan. Dr. Navabi, who studied for some time 
at the School of Oriental and African Studies, very generously 
allowed me the use of his field notes, as well as of a fair copy 
he had made of the larger part of his collections. Naturally, 
the publication of this material, which is fairly comprehensive, 
must be left to Dr. Navabi ; but he has kindly permitted me 
to quote from it in this paper. 

The dialect of Galin-qaya has inevitably been strongly 
influenced by Turkish, which is also the second language of its 
speakers. There is a good deal of vowel assimilation : umiita 
' to teach ' corresponds with Persian amuytan, vbrbr- ' to pass 
by ' with Persian guhar- ; in zunusna, vbrosna, beramesna, 
respectively ' he knows ', ' it rains ', and ' he cries ', the vowel 
in the penultimate is one and the same by origin. The palatal- 
ized ^-sounds are so strongly marked that they appear to be 
scarcely distinguishable from ^-sounds ; thus in Dr. Navabi's 
notes one finds both Hina and cina for ' girl ', Holla and cblla 
for ' you made '. A preceding genitive is mostly resumed by 
the enclitic pronoun for the 3rd person, which is -y after 
a vowel, j after a consonant, and -y)- between vowels ; e.g. 
Hold = ' hat ', yan = ' wife ' : ' his hat ' is Jcblby, ' his wife ' 
is yan), and Hblbyja means ' it is his hat ' (the -a at the end 
being the word for ' is '). This pronoun now appears where 

1 The 3rd edition of Kesravi's Ahari (see above p. 166, n. 2) also contains 
specimens of the zaban-i Harzand, 17 everyday sentences (pp. 63-64) and 
a brief word-list (pp. 62-63). Although owing to the orthography used 
(unvocalized Arabic script with somewhat haphazard matres lectionis) some 
points necessarily remain uncertain, one can say that the language of these 
specimens is substantially in agreement, indeed almost identical, with that 
of Dr. Navabi's collections. Curious is the uncertainty in the endings of the 
1st pers. plur. (also of the 1st pers. sing.), e.g. subj. him ' let us go ' (Gal. 
annum) ; pres. (n)kwndwm ' we (do not) do ' (Gal. kondum), but (n)znswn 
' we (do not) know ' (Gal. zunusnum ; prob. misprinted for *nznsnwn) ; 
pret. (hth) brum ' we were (asleep) ' (Gal. heta berum) ; " trans." pret. 
zwnwsm'n' wrtmwn ' we knew (and) fled ' (Gal. -muna) ; uncertainty in this 
very point is found also in Dr. Navabi's notes. 



W. B. KENNING — THE ANCIENT LANGUAGE OF AZERBAIJAN 169 

a genitive precedes, e.g. merde koloy ' the man's hat ' or brori 
yan) ' the brother's wife ' — a construction doubtless due to 
Turkish influence. Similarly, the possessive pronouns {caman, 
esde, avey ; cama, sema, avuney), which incidentally resemble 
the Talis! and Takistani forms, are resumed by enclitic 
pronouns, e.g. caman yanma ' it is my wife ', esde zunustar 
' your knowledge ' (the final -r being the enclitic pronoun of 
the 2nd person). 

Here I have put down a scheme of the nominal inflexion, in 
which, however, one or two points are not entirely certain : — 



General Case 


Sing 


. merd 


Plur. yanoy 


Oblique Case 




merde 


yanun 


Determ. Ace. 




merde 


yanune 


Locative 




merdenda 


(yanunenda) 


Ablative 




merderi 


yanuneri 


Instrumental 




merdehun 


(yanunehun) 


Partitive 




— 


yanunda 



It is always interesting to see how a language that has once 
shed an elaborate system of inflexion creates a fresh system 
that bears comparison with the original one. Actually, most 
of these cases are pseudo-cases, mere juxtaposition of noun 
and postposition. That becomes clear when an enclitic pronoun 
is affixed to the noun ; it precedes postpositions, but of course 
follows proper endings. Hence dosmonda ' in my hand ', 
doslonda ' in your hand ', karjiri ' from his house ' (kar 
= house), karlenda ' in your house ', hamaypri ' from all of it '. 
This consideration shows also that the determinative accusa- 
tive, which in the singular often coincides with the oblique 
case, is in fact a pseudo-case. Actually, it is distinguished when 
a noun ends in a vowel ; then the accusative ends in -re, but 
the oblique case is identical with the general case ; e.g. Jcina 
' girl ', oblique case also Jcina, but liinare is the determinative 
accusative. Now I have found two examples in which enclitic 
pronouns are involved, grlle ' his neck ', i.e. giri ' neck ' -j- -y- 
for the 3rd person + the dissimilated accusative ending ; and 



170 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 



xoyo esde Jcarille vorun Jcani ' may God ruin your house ', where 
karille ' your house ' is composed of Icar ' house ', the enclitic 
pronoun of the 2nd person (-r, often -I), and the mark of the 
accusative. 

It is not possible to discuss here all the problems connected 
with the verb of this dialect. The following table contains the 
principal forms of the preterite and the perfect : — 

I saw I have seen I went I have gone I have not gone 



vimma 


vindama 


serin 


serama 


nema sera 


villa 


vindara 


seri 


serara 


nera sera 


vinja 


vindaya 


Sera 


seraya 


niya sera 


vimmuna 


vindamuna 


serum 


seramuna 


nemuna sera 


vinnura 


vindaruna 


serur 


seranura 


nenura sera 


vinjuna 


vindayna 


serut 


serayna 


neyna sera 



It shows the influence of the transitive verb upon the intran- 
sitive. The perfect of the latter (serama) is entirely modelled 
on that of the former (vindama) ; similarly behesdama ' I have 
got up ', nehetara ' you have not slept ', vin nani ziilferangin 
gune parisun berama . . . hazrate Adam angin oso pasimun 
berama ' see, how like your locks I have become tangled in 
(my) soul (?)... like Adam I have now become repentant '. l 
The intransitive preterite ordinarily preserves its distinct 
forms (Serin), even where the two kinds of verbs are in close 
contact, e.g. serim nahare horma amarim 2 ' I went, ate lunch, 
came ' ; but occasionally we find forms adapted to the transi- 
tive, e.g. vororma ' I passed by ', morja ' he died '. 

Of particular interest are the verbal stems, of which a fairly 
full list is given here. In Middle Iranian and in the majority of 
modern Iranian languages we find two verbal stems, a present 
stem and a preterite stem ; but in the dialect of Galin-qaya 
there exist three stems, present, preterite, and subjunctive, 
and an additional form for the 2nd singular of the imperative : 

1 There is no translation of these verses in Dr. Navabi's notes ; the one 
above is therefore conjectural, -angin ' like ' (spa vorgangin a ' a dog is like 
a wolf ') represents earlier *angon, of. MPers. hngum, Sogd. 'nyum. 

2 With regard to -im instead of -in, see above, p. 168, n. 1. 



W. B. HENNING — THE ANCIENT LANGUAGE OP AZERBAIJAN 171 





Imp. 


Subj. 


Pret. 


Pres. 


Base 


(a) 


se 


sun- 


ser- 


send- 


syaw 


(b) 


-- 


bun- 


ber- 


— 


baw 


(c) 


cZa 


dan- 


dor- 


dond- 


da 


(d) 


lea 


kan- 


Jcord- 


kond- 


kar 


(e) 


(bi)a 


(bi)an- 


ord- 1 


— 


d-bar 


(f) 


— 


han- 2 


hord- 


— 


x w ar 


(g) 


«ierar 


— 


vorord- 


_— 


wi-tar 


(h) 


-— 


— 


vozord- 3 


vdzorn- 


wi-car 


(i) 




yan- 


yar- 


yand- 


jan 


0') 


dsta 


astan- 


dstar- 4 


dstand- 


d-stan 


(k) 


vin 


vin- 


vind- 


vinn- 


wain 


(1) 


- ■ 


— 


[hond-] 5 


honn- 


x w an 


(m) 


.._ 


zun- 


zunust- 


zunusn- 


zan 


(n) 




os- 


ot- 


otn- 


wak 


(o) 




heras- 


horot- 


horodd- 


fra-wak 


(P) 


n 


— 


ret- 6 


rett- 


raik 


(q) 


— 


duz- 7 


— - 


dutn- 


dauk 


(r) 


—- 


hes- 


het- 


hett- 


x w ap 


(8) 


gen 


gen- 


got- 


— 


grab 


(t) 


[nsm] s 


— 


nest- 


nesn- 


ni-had 


(u) 


— 


fi- 


fest- » 


fesn- 


vid' ? 


(v) 


[bend-] 10 


— 


bost- 


bosn- 


band 


(w) 


— 


— 


nevest- 


nevesn- 


ni-pais 


w 


-- 


— 


host- 


hosn- 


x w az 


(y) 


be 


bez- 


beliest- 


behesn- 


bar + haiz 


(z) 


— 


— 


— 


vasn- n 


waxs 



1 Preterite stems ending in two consonants are shortened in the " tran- 
sitive " preterite, generally by the loss of the second consonant ; hence in 
Kesravl's material 'wrf = orja ' he brought ' ; zwnwsl' — zunusla ' you 
knew ' ; cf. above vimma, villa, etc. (from vind-ma, vind-la etc.). 

2 In Kesravl's material hynm as 1st sing. Subj. (read hnm ?). 

3 ' To perform ' (a prayer, etc.), Pers. guzardan. 

4 ' To buy.' 6 Supplied from Kesravi (hwnd-). 
6 Cf. veretu (inf.) ' to flee ' ; Kesravi wrlmwn (above p. 168, n. 1) = veret- 

muna. ' ' To sew.' 8 Supplied from Kesravi. 

9 ' To throw, spread (a rug over someone) ' and ' to scatter seed, to sow \ 
Pers. equivalents andap(tan and gustardan. Example, artandahioy noxut 
festayna ya neyna festa ' have the people there sown (chick-)peas or not ? ' 
(aria ' thither ', arteyri ' thence ', artanda ' there ', artandaki ' (someone) 
being there ', artandahioy pi). 

10 Adapted from Lehmann-Haupt (bind-, above p. 167). 

11 ' To shine, be alight ' (osma vasna ' the moon is shining ', core vasna 
' the lamp is lit '). 



172 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 

The preterite stem continues the ancient preterite stem and 
thus presents no difficulty. The old present stem is continued 
substantially by the subjunctive stem, and entirely by the 
imperative, which generally has strongly shortened forms. 
However, some of the subjunctive stems, those at the beginning 
of the list (a-f ), have acquired a final -n, the origin of which 
is not clear. Here only those stems are affected, the base of 
which ended in a vowel or an unstable consonant. It is prob- 
able that these subjunctive stems are new formations built 
upon the imperative ; for example, case (d), ' to make ', 
where the old present stem was kar, which in the imperative 
was shortened to lea : on this form the subjunctive stem was 
built by the addition of -n-. The first verb in the list shows 
that this formation is not entirely recent ; for the subjunctive 
stem sun- must have been created at a time when the im- 
perative was still *su. 

The most interesting of the three stems is the present stem. 
It is evidently built on the preterite stem ; in this point the 
dialect of Giilin-qaya differs from most Western Iranian 
languages. This origin of the present stem is quite clear in the 
forms in the second half of the list ; but some of those at the 
beginning of the list are again difficult. For example, if we 
consider case (i), at first sight one might assume that the 
present stem, yand-, was derived from the subjunctive stem, 
yan-, rather than from the preterite stem, yar-. However, it 
is in itself unlikely that in the weaker bases the formation 
should have been essentially different from that found in the 
stronger ones ; also, it is to be observed that the vowel of the 
present stem is in every case identical with the vowel of the 
preterite stem. 

The rule for the formation of the present stem can be formu- 
lated in this way : if the Old Iranian preterite stem ended in 
-I- preceded by a sonant, then the Galin-qaya present stem 
ends in -nd with loss of the old -t- ; and if the Old Iranian 
preterite stem ended in -t- preceded by a consonant, then the 
Galin-qaya present stem ends in that consonant, if it was 
preserved, plus -n-, otherwise in -tn or -tt. The origin of these 



W. B. HENNING — THE ANCIENT LANGUAGE OF AZERBAIJAN 173 

stems may be found in the -ant- participle, which in Galin-qaya 
is regularly built on the preterite stem : seranda ' going ', 
doranda ' giving ', kordanda ' making ', yaranda ' hitting ', and 
so forth. These forms may have been strongly shortened in 
the creation of the present stem, so that the actual present 
would be a composite tense by origin, ' I am going ' in the 
place of ' I go '. Thus one could account for the various finals, 
-nd-, -n-, and -d- (often assimilated to -t-), all resulting from 
-nd- at the end of clusters of consonants. 

These few details will, I hope, give an idea of the distinctive 
features of the dialect of Galin-qaya, which we may regard as 
representative of the Harzanl group. The question now arises : 
is this Harzani the last surviving form of the language once 
spoken in Azerbaijan 1 This question is not easily answered. 
The first test to be applied concerns certain sound-changes 
which must be presumed to have characterized the old 
language of Azerbaijan, principally the change of Iranian fr- 
to hr-, and the change of intervocalic -d- to -r- ; the Iranian 
loanwords in Armenian, which entered Armenian from the 
neighbouring province of Azerbaijan, prove the existence of 
these changes. Now the first, hr from fr, is satisfactorily 
present in Harzanl ; but it carries no weight, because it is also 
present in at least half the North-western Iranian languages. 
The second change, -r- from intervocalic -d-, is unfortunately 
absent. It is true, intervocalic and post-vocalic -t- regularly 
becomes -r- in Harzani ; many of the words I have quoted 
show it — but that should not be confused, though it often has 
been confused, with the change we are seeking. Intervocalic 
-d- has either disappeared in Harzani, or been replaced by a 
glide. There is only one certain case of -r- from -d-, arina 
' Friday ', and that word is probably a loan-word ; a doubtful 
case is the word for ' under ', which occurs in ruzare ' west ', 
literally ' sun-down ' (ruz ' sun '), and in parare ' below ', 
which contrasts with parpe ' above ', so that -are meant ' under ' 
and may be referred to Old Iranian adari ' under ' : that word, 
however, already contained an -r-, so that we are not safe in 
claiming that the Harzani -r- in this word represents the old -d-. 



174 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 

We have now to consider the relationship between HarzanI 
and the other languages of the north-western group. It is 
obvious, and has already been pointed out by Dr. Baer, 1 that 
HarzanI is most closely related to Talisl. This relationship 
would be even stronger if Talis!, which now presents much 
abbreviated forms through the loss of interior -r-, all dentals, 
and other consonants as well, had once shared in the change 
from intervocalic -t- to -r-, which characterizes HarzanI ; it 
has indeed been reported that words exhibiting that change 
occur in the southernmost dialect of Talisl, that spoken in 
Assalim. 2 It should be noted that Talisl, like HarzanI, 
possesses a present built on the preterite stem ; the Talisl 
forms have not been explained correctly, 3 and can in fact be 
explained only with the help of the HarzanI material. 

On the other side, HarzanI is related to Zaza. One may 
instance the existence of two genuine plural cases in both 
languages ; the ablative postposition -ri, Zaza -ra ; the 
negative prefix in cini(ya) i ' it is not ', Zaza cinyo, cinya I 
many characteristic words, such as HarzanI iisma ' moon ', 
Zaza a&ma, the nearest related word 5 being Talisl ovsim ; vasna 
' it shines ', Zaza vasena ; gen- both ' to take ' and ' to fall ' in 
HarzanI and Zaza ; rau ' quick ' in both languages 6 ; vondor- 
' to stand ' : Zaza vindar- 7 and Vafsl vender-, 8 a verb known 

1 Anderseits erweisen sich das Harzdndi und das Shalwudl Azdrbaijans mi 
dem Talyshi Kaspiens eng verwandt (Actes, pp. 155 sq.). 

2 B. V. Miller, Tal'isskiy Yaz'ik, 1953, p. 261. Two of the words quoted 
above, p. 160, n. 2 (sr and brh) prove that this change als ooccurred in the 
Khalkhal dialect from which Kesravi's specimen is derived. 

3 Thus B. V. Miller in his latest work explained the Tal. present (votedam, 
or shorter vottam, ' I say ') as compounded of the infinitive (vote), the pre- 
position da' in' (used as postposition), and the present of ' to be ' (TaVisskiy 
Yaz'ik, p. 146). However, Miller himself does not attribute, in his discussion 
of the prepositions (ibidem, pp. 86-88), such a preposition to Talis! (only di 
' with ', and the postposition ada ' in ', which would not produce the form 
required ; there is, of course, a preverb da). 

4 Several times also in Kesravi's specimen, spelt iynyh. 

5 | Now : Kerlngani uarnu.] 8 [So also in Kermgam.] 

7 Hadank, Zaza, 138, 273, 361, 378, ste.hen bleiben, stillstehen, anlialten, 
bleiben, warlen. Similar forms (but with initial m-) occur also in Gurani. 

8 Muqaddam, Vafs, p. 97, inf. venderdan, imp. hawender, pret. hawcnderd, 
and similarly in neighbouring villages (note also vandarda, p. 127 middle) ; 
Pers. equivalent istadan. [Add KeringanI vendardan.] 



W. B. HENNING — THE ANCIENT LANGUAGE OF AZERBAIJAN 175 

from Middle Iranian. 1 This is merely a small selection of the 
many coincidences, some of them exclusive, that can be 
quoted. They leave no doubt that Harzani takes its place 
between Talis! and Zaza. 

This result seems to be consonant with the present geo- 
graphical location of the three languages, with Harzani 
actually in the middle between TalisI in the east and Zaza in 
the far west. But these appearances are deceptive ; for it is 
certain that Zaza, now altogether out of contact with the 
languages to which it is related, has been carried to Eastern 
Turkey by some migration. It has been argued on historical 
and linguistic grounds, on converging lines, that Zaza is a 
branch of the ancient language of Dailam, an alpine country 
in the centre of the high mountains on the southern shore of 
the Caspian Sea. 2 If Zaza had its original place in Dailam (to 
the north and north-east of Takistan), we are driven to the 
assumption that Harzani, too, is a dislocated language and 
had its home to the south-east of Talis, somewhere between 
Talis and the ancient Zaza country. Such an assumption would 
give an answer to many difficult problems ; for example, we 
should gain a perfect series of the languages that form their 
present stem with the help of an -n- or -nd- suffix, that is, 
TalisI, Harzani, Zaza, parts of GilakI, Tabari, and some 
dialects near Samnan. 3 Moreover, there has been a report, by 
the head of the American Presbyterian Mission in Tabriz at 
the beginning of this century (S. G. Wilson), that the people 
of Harzan had been transferred there from the region of Talis 
by Nadir-Sah, that is to say a little over two hundred years 
ago. 4 Ordinarily one might look upon such a report with 

1 Pahlavi Psalter wndlty = qayyama ' durans, permanent ' ; Inscr. of 
Shapur, Parthian 17 RB' y'ztn wyndrsn 'BDt ; with assimilation -nd- 
> -nn- Manieh. MPers. wnyr- ' to remain (permanently), stay ' (so to be 
translated), from which its apparent causative wyn'r- ' to set, fix ' (also 
Pahlavi wyril-, Pahl. Ps. wrilty, Inscr. wri'l-) cannot easily be separated. 
The derivation of wnyr- from a base nar- (cf. Zll., ix, 206 ; Bailey, JRAS., 
1953, 106) can scarcely be maintained. 

2 See the full discussion in the introduction to Hadank, Zaza ; cf. 
Minorsky, Domination des Dailamites, 17 ; BSOAS., xi (1943), 86-89. 

3 Cf. Hadank, ibidem, p. 23. * See Hadank, ibidem, p. 5. 

o 



176 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1954 

a critical eye 1 : but here, as it is in agreement with the 
linguistic evidence, we may accept it as corroboration. 

That leaves us with empty hands. The dialects that were 
presumed to be the last remnants of the ancient language of 
Azerbaijan have proved to be t recent imports from another 
province. We are similarly unfortunate with the literary 
evidence which the late Kesravl had collected from Persian 
works : there is nothing conclusive. The most considerable 2 
is a set of 14th century dialect quatrains from Ardabil 3 ; but 
Ardabll is merely on the fringe of Azerbaijan, close to Talis, 
and the dialect of these quatrains has been shown to be akin 
to Talisl. 4 In short, with all the dialects we have considered 
here, of Talis, Harzan, Khalkhal, and Takistan, we remain on 
the threshold of Azerbaijan ; but of the language once spoken 
in Azerbaijan itself we know nothing. 5 

Additional Note 
This paper has been dogged by misfortune. Important new 

1 For example, I was myself told in Takistan that according to a local 
tradition the people of that village (the Tat) were immigrants from somewhere 
else ; but no one could say from where they had come, or at what date. 

2 The 3rd edition of Kesravi's Ahari contains additional dialect verses 
(some of them very interesting) taken from various fangs (pp. 54-59) ; 
unfortunately, they are not sufficiently closely localized. 

3 Silsilatu 'nNasab-i Safawvye, pp. 29-32 ; Kesravi, Ahari, 1st ed., 
pp. 31-42 (3rd ed., pp. 36^6). 

4 See Miller, Tal. YazVc, 254-263. After seeing Miller's work (1953) 
I abandoned my intention to give a full analysis of the language of these 
quatrains. Note that drd{h)-zr continues Manich. Parthian drdjd ' suffering, 
in pain ' (cf. srmjd) ; and zyr ' life ' Manich. Parth. jyd. There is a possible 
case of -d- > -r- in gwrym (6) ' I am the ball ', which reading, however, is 
secured only by emendation (required by the rhyme) ; nwstym (7) is mis- 
reading of n-dstym, cf. Tal. daze Miller, Tal. Teksli, 217 ; Giilin-qaya 
dosdom dasna ' my hand hurts ' ; probably Yaghnobi daxs-. The word for 
'God', 'wy'n, is surely of Turkish origin (= oyan Kasyari ; Houtsma, 
Glossar, p. 7 ; etc.). 

6 The ' Appendix ' to the Risale by Kuhl of Anarajan (16th century), to 
which A. Iqbal has drawn attention (Yddgar, ii, fasc. iii, pp. 43-50) and 
which was published in full by M. Muqaddam ( Yak sanad-i tarixi az guyU-i 
ahari-i Tabriz = Iran-Kude 10, 1317 Yezd./1948), is not, as has been 
claimed, written in any " dialect ", but in a vulgar (extremely vulgar !) and 
slangy type of colloquial Persian. It has no bearing on the problem of the 
ancient language of Azerbaijan. 



W. B. HENNING — THE ANCIENT LANGUAGE OF AZERBAIJAN 177 

publications, each requiring fresh adaptation of the views 
presented, kept coming in, either at the last minute or after 
the event. While the printing was in process, my friend and 
former student Dr. E. Yarshater, of the University of Tehran, 
presented me with (1) the second edition of Kesravl's ASari, 
for which I had looked so long in vain (above, p. 165 ; it is 
indeed much superior to the third edition, which is merely a 
reprint) ; (2) M. Sutude (Sotoodeh), Farhang-i GilaJci, Tehran, 
1954 (an admirable work) ; (3) Yahya Zoka, The dialect of 
Keringan, Tehran, 1332/1953-4 (the first glimpse of one of the 
unknown dialects of the Qaraja-day, see above, p. 165 ; 
received too late, unfortunately, to be taken into account 
here).